AI Will Transform Public Safety. Our Responsibility Is to Make Sure It Transforms It Responsibly.

By Jeff Robertson
President, Allerium

Artificial intelligence will transform public safety. In time, it will likely touch nearly every part of emergency communications, from how calls and data are processed to how incidents are understood, how telecommunicators are supported and how agencies operate.

Some of that transformation is already underway. AI can transcribe and translate conversations, summarize information, automate repetitive tasks and support quality assurance and training. Other applications are emerging quickly, while many of the possibilities being discussed today simply aren’t ready yet.

That distinction matters. The question facing public safety isn’t whether AI is coming. It is how we prepare for it, where we use it and how we make sure the technology serves the mission rather than simply adding another layer of complexity.

At Allerium, we believe the industry should move fast on understanding AI and move deliberately on deploying it.

Start with the problem, not the AI

“AI” has become a blanket term for technologies that vary enormously in what they do and the risks they introduce. Simple automation, machine learning, generative AI and systems capable of taking actions are routinely discussed under the same label. They shouldn’t be evaluated the same way.

Before asking where to use AI, public safety agencies should start with a more fundamental question: What problem are we trying to solve?

That might be repetitive work consuming a telecommunicator’s time, important information buried across multiple systems, supervisors able to review only a fraction of interactions, or simply too much information arriving for any one person to reasonably process.

NG9-1-1 makes this last challenge particularly important. The industry has spent years building the infrastructure to make richer information available, but more information does not automatically mean better situational awareness. If every new source creates another feed, alert or screen, we haven’t solved complexity. We have moved it.

This is where AI has enormous potential: not simply to give people more information, but to help determine what information matters.

Think beyond the call

Much of today’s conversation about AI in public safety focuses on the telecommunicator, and for good reason. During an emergency interaction, AI can already support applications such as transcription and translation, while emerging capabilities could help recognize information already provided and prompt for details that may be missing. Afterward, AI can also support quality assurance, coaching and training by helping agencies review interactions and identify patterns.

But some of the most important applications of AI may happen before information ever reaches the console.

A major incident can generate multiple calls across multiple PSAPs alongside vehicle information, video, sensors and other authorized data sources. Each piece may tell part of the same story. Intelligence applied across the emergency communications ecosystem could help recognize those relationships, reduce duplication, identify what has changed and surface relevant information before another stream of raw data reaches a telecommunicator.

That requires more than AI. It requires structured, standards-based incident information that can move across systems and agencies with appropriate security, permissions and controls while preserving context. AI can then help turn that information into relevance.

The goal isn’t to put everything we know in front of a telecommunicator. It is to get the right information to the right person at the right time.

Take work off people. Don’t take people out of the loop.

As AI becomes more capable, the conversation inevitably turns to what it means for the people doing the work today. We think the distinction is important.

The goal should not be to preserve every task simply because a person performs it today. AI should take on repetitive work where automation makes sense. It should process information at a scale people cannot. It should help reduce cognitive burden in an environment where attention is incredibly valuable.

But taking work off people is different from taking people out of the loop.

AI can analyze, correlate, summarize, translate and recommend. People bring context, experience, judgment and the ability to communicate with someone who may be experiencing the worst moment of their life. How that balance works will change as the technology improves, and it will vary depending on the consequence of the task.

The more consequential the action, the more carefully we should think about where people remain in the process.

Innovation should strengthen public trust

Public safety has earned a level of trust that few institutions enjoy. As agencies begin exploring AI, there is an opportunity to carry that trust forward.

Historically, public safety has tended to adopt technology only after it has reached a high level of maturity, stability and reliability. At Allerium, we describe that standard as public safety grade. AI is changing that pattern. The technology is evolving rapidly, and some agencies are already testing and evaluating new capabilities as they develop.

We believe that shift is important. Public safety doesn’t need to wait on the sidelines while AI evolves. Agencies should have a role in testing, learning and helping shape how the technology is ultimately used in emergency communications.

But that leadership comes with a responsibility to bring the public along. People place enormous trust in 9-1-1 and emergency services, even as broader public confidence in AI is still developing. Agencies and their technology partners should be able to explain where AI is being used, why it is being used and what safeguards surround it.

Responsible adoption gives public safety an opportunity to do more than protect the trust it has earned. It can help demonstrate what thoughtful, mission-driven use of AI looks like.

Your AI strategy is also a data strategy

Perhaps the most important questions about an AI system aren’t visible in the demo.

Emergency communications contain extraordinarily sensitive information: calls, recordings, transcripts, location information and incident details. Agencies need to understand not only what an AI system can do with that information, but what happens to the information itself.

Where does the data go? Who owns it, who controls it and what rights does each party have to use it? Who can access it? Is it retained? Can it be used to train a model or for another purpose? Does it pass through another company, cloud provider or AI platform? What happens to it when the relationship ends?

The same questions apply when an AI capability is offered at little or no cost. Free does not necessarily mean there is a problem, but agencies should understand the exchange. In some cases, access to data or rights to use that data may be part of the value received by the provider. An agency may decide that trade-off is acceptable, but it should be explicit and understood rather than assumed.

These aren’t simply technology questions. They are questions agencies should be asking when evaluating the partner, the architecture and the contract behind the AI.

Public safety has unique responsibilities and authorities because lives can be at stake. That should make careful data governance more important, not less. Specific legal, regulatory, records-retention and privacy requirements will vary by jurisdiction, and the landscape surrounding AI is evolving quickly. Agencies should understand how those requirements apply before introducing AI into the workflow and work with partners that are prepared to keep pace as the rules continue to change.

It also means agencies need to understand the partners behind the technology. A vendor may rely on third-party models, cloud services and data processors to deliver an AI capability. There is nothing inherently wrong with that, but a trusted public safety partner should understand the full chain and be able to explain it.

A good AI demo should answer what the technology can do. A good AI partner should also be able to answer what happens behind it.

Data ownership and control, security, retention, model training and permitted uses shouldn’t be discovered after deployment. They should be understood before it and, where appropriate, defined in writing.

We are all still figuring this out

For all the excitement surrounding AI, there is a reality the industry should be comfortable acknowledging: none of us has the complete roadmap.

Technology companies are learning. Agencies are learning. Regulators and standards organizations are learning. Allerium is learning too. The capabilities will change, and our understanding of where they belong will change with them.

That isn’t an argument for waiting. Public safety should experiment, test and learn, while recognizing that how it does so can affect the trust communities place in emergency services. Agencies should ask difficult questions, measure results and understand where AI performs well and where it doesn’t. Vendors should be transparent about what their technology does, what happens to customer data and where the limitations are.

AI will transform public safety. The organizations that benefit most won’t necessarily be those that adopt the most AI the fastest. They will be the ones that understand what they are trying to accomplish, establish the right safeguards and use the technology where it genuinely makes public safety better.

Move fast on understanding AI. Move deliberately on deploying it.